The browser as a shared canvas

Frisson Labs is building a technical architecture that treats the web browser not just as a document viewer, but as a space for interactive, embodied AI. While most AI companies focus on the intelligence of the model or the efficiency of the task, Frisson is preoccupied with the concept of presence. Their lead product, Oto, is a character that lives inside Chrome, designed to interact with the websites a user visits. This is an attempt to move digital companionship beyond the ephemeral chat window and into the actual activities of a user's digital life, such as watching YouTube videos or co-working.

Building this requires more than just a wrapper around an LLM. The company has developed a custom web-based game engine designed to handle real-time 3D rendering through WebGL and Three.js. This engine allows their companions to exist as modular character systems that can scale from 2D sprites to high-fidelity 3D models. The engineering challenge they are tackling involves creating physics engines that understand arbitrary DOM structures. This means an AI companion can theoretically recognize and interact with the specific buttons, images, and text on any given website without breaking the underlying functionality of the page.

Technical art and engine design

The team is led by co-founder Charles Niu and operates out of San Francisco with an in-person culture. The technical background of the founding team suggests a focus on the bridge between high-end game engineering and web performance. They are hiring for roles that emphasize WebGL shaders and asset pipelines, specifically looking for ways to run over a hundred simultaneous AI companions without degrading the performance of the host machine. This focus on optimization is a necessity for their vision of a "shared experience engine" where AI and humans occupy the same visual space.

Frisson Labs identifies more with the culture of gaming and VTubing than with traditional SaaS or enterprise AI. Their recruitment of UGC creators and social media specialists indicates a strategy built on viral, creator-native content. They aim to replicate the appeal of popular AI characters like Neurosama, but through a more accessible, browser-native format. By making these companions interactive and aware of the user's context, they are attempting to solve the isolation inherent in current AI interactions.

Funding and market position

The company is backed by General Catalyst and South Park Commons, two organizations that typically bet on high-conviction, technically heavy consumer plays. In a market saturated with productivity-focused agents, Frisson Labs is an outlier focusing on the emotional and social utility of AI. They are betting that as LLMs become a commodity, the differentiator will be the user interface—specifically, how these models are embodied and how they share space with the user. The goal is to create a digital companion that users "cherish," leaning into the parasocial dynamics that have already been proven out in the streaming and gaming industries.